U.S. Patent: 6261914 - Process for improving local uniformity of chemical mechanical polishing using a self-aligned polish rate enhancement layer

Back to EveryPatent.com

United States Patent	*6,261,914*
Divakaruni , et al.	July 17, 2001

Process for improving local uniformity of chemical mechanical polishing using a self-aligned polish rate enhancement layer

Abstract

A method for making a semiconductor device, includes forming an oxide layer on a silicon substrate, forming a nitride layer over the oxide layer; depositing one of a doped oxide layer and an undoped porous oxide layer on the nitride layer, etching trenches through the one of the doped layer and the undoped porous oxide layer, the nitride layer, and the oxide layer, depositing an undoped oxide layer to fill the trenches, and patterning the undoped oxide by chemical mechanical polishing (CMP).

Inventors:	Divakaruni; Ramachandra (Middletown, NY); Gambino; Jeffrey Peter (Westford, VT); Radens; Carl J. (LaGrangeville, NY); Stephens; Jeremy K. (Wappingers Falls, NY)
Assignee:	International Business Machines Corporation (Armonk, NY)
Appl. No.:	361961
Filed:	July 27, 1999

Current U.S. Class: 438/359; 257/E21.244; 257/E21.548; 438/360

Intern'l Class: H01L 021/331

Field of Search: 438/359,361

References Cited U.S. Patent Documents

4962064	Oct., 1990	Haskell et al.
5192706	Mar., 1993	Rodder.
5362669	Nov., 1994	Boyd et al.
5459096	Oct., 1995	Venkatesan et al.
5494857	Feb., 1996	Cooperman et al.
5516625	May., 1996	McNamara et al.
5665202	Sep., 1997	Subramanian et al.
5691215	Nov., 1997	Dai et al.
5728621	Mar., 1998	Zheng et al.
5733801	Mar., 1998	Gojohbori.
5786262	Jul., 1998	Jang et al.
6017803	Jan., 2000	Wong	438/430.
6150212	Nov., 2000	Divakaruni et al.	438/244.

Primary Examiner: Booth; Richard
Assistant Examiner: Gurley; Lynne A.
Attorney, Agent or Firm: McGinn & Gibb, PLLC

Claims

What is claimed is:

1. A method of making a semiconductor device, comprising:

forming an oxide layer on a silicon substrate;

forming a nitride layer over said oxide layer;

depositing an enhancing layer comprising one of a doped oxide layer and an undoped porous oxide layer, on said nitride layer;

etching trenches through said enhancing layer, said nitride layer, and said oxide layer;

depositing a selectively enhanced layer comprising an undoped oxide layer to fill the trenches; and

patterning said selectively enhanced layer by chemical mechanical polishing (CMP).

2. The method according to claim 1, wherein, with said doped oxide layer, the CMP is stopped before the doped oxide layer is removed from regions with predetermined wide active areas.

3. The method according to claim 1, wherein with said porous oxide layer, the CMP is stopped when the nitride layer is exposed in all regions.

4. The method according to claim 2, further comprising:

stripping said doped oxide layer selectively to said undoped oxide layer, stripping said nitride layer, and stripping said undoped oxide layer, and growing a sacrificial oxide layer for subsequent implants.

5. The method according to claim 1, wherein said doped oxide layer comprise doped SiO.sub.2.

6. The method according to claim 1, wherein said selectively enhanced layer comprises polysilicon.

7. The method according to claim 6, wherein said enhancing layer comprises doped polysilicon, and said selectively enhanced layer comprises intrinsic polysilicon.

8. The method according to claim 1, wherein said oxide layer has a thickness in a range of about 2 nm to about 50 nm.

9. The method according to claim 8, wherein said oxide layer has a thickness of about 10 nm.

10. The method according to claim 1, wherein said nitride layer comprises a layer of Si.sub.3 N.sub.4 deposited by Low Pressure Chemical Vapor Deposition (LPCVD) to a thickness of substantially in a range of about 50 nm to about 300 nm.

11. The method according to claim 10, wherein said nitride layer has a thickness of about 100 nm.

12. The method according to claim 1, wherein said doped oxide layer comprises a doped SiO.sub.2 layer deposited to a thickness in a range of about 50 nm to about 300.

13. The method according to claim 12, wherein said doped oxide layer has a thickness of about 100 nm.

14. The method according to claim 1, wherein said porous oxide layer comprises a porous SiO.sub.2 layer deposited to a thickness of about 50 nm to about 300 nm.

15. The method according to claim 14, wherein said porous oxide layer has a thickness of about 100 nm.

16. The method according to claim 1, wherein said trenches are etched to have a depth of about 100 to about 1,000 nm.

17. The method according to claim 16, wherein said trenches have a depth of about 200 nm.

18. The method according to claim 1, wherein said undoped oxide layer comprises an undoped SiO.sub.2 layer deposited to fill the trenches to a thickness of about twice a depth of a trench, and

said method further comprising a high temperature anneal for selectively driving impurities into the undoped SiO.sub.2 layer from the doped oxide layer, thereby resulting in self-aligned doping of said undoped SiO.sub.2 layer.

19. The method according to claim 1, further comprising:

annealing said device to drive impurities from said enhancing layer into said enhanced layer to form self-aligned doped portions of said selectively enhanced layer.

20. The method according to claim 19, wherein a chemical mechanical polishing (CMP) rate is higher in said self-aligned doped portions than in undoped portions of said selectively enhanced layer.

Description

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention generally relates to a method for producing a semiconductor device, and more particularly to a method for improving uniformity of a chemical mechanical polishing (CMP) operation used in producing the semiconductor device.

2. Description of the Related Art

Conventional systems utilize shallow trench isolation (STI) in advanced integrated circuits to electrically isolate neighboring devices. Chemical mechanical polishing (CMP) is often used to pattern insulators in semiconductor trenches or conductors in insulating trenches. In many applications, the polishing is stopped only after all the material has been removed from the field regions.

However, oftentimes this operation results in over-polishing of the material in wide trenches.

For example, FIGS. 1A-1D illustrate a conventional process which is relatively simple, and is extendable to sub-quarter micron dimensions. In FIG. 1A, a pad oxide 11 is grown by thermal oxidation of a silicon substrate 10. Thereafter, a pad nitride, formed for example of SiN 12, is deposited on the SiO.sub.2 layer. Then, the trenches are etched.

In FIG. 1B, a SiO.sub.2 layer 14 is deposited to fill the trenches.

In step 1C, the SiO.sub.2 layer 14 is patterned by CMP.

Finally in step 1D, the pad nitride and pad oxide are stripped, and a sacrificial oxide 15 is grown.

However, a problem with the conventional process is that the polishing process used to pattern the SiO.sub.2 isolation, as shown in FIG. 1C, may result in localized variations in the SiO.sub.2 thickness. This is a problem, and could result in the resulting device "failing" during operation.

That is, in regions where the trenches are wide or the density of the active area is low, such as in region C of FIG. 1D, there is excessive thinning of the SiO.sub.2. The excessive oxide thinning allows the gate to wrap around the active areas, resulting in a low threshold voltage for the affected devices. Again, this may cause the device(s) to fail.

Moreover, in regions where the active areas are wide or where there is a low density of trenches, there may be insufficient removal of the SiO.sub.2 such as in region A in FIG. 1D. The residual SiO.sub.2 masks the pad nitride strip, and the pad nitride blocks subsequent implants and gate oxide growth, resulting in failing devices.

Localized variations in polishing resulting from variations in the pattern factor are also observed for other CMP processes. For example, such processes include metal CMP to produce damascene interconnects (e.g., see FIGS. 3A-3C described below) or studs, dielectric planarization over a gate stack (e.g., see FIGS. 5A-5C described below) or a metal stack, and polysilicon patterning in deep trenches.

In the case of metal CMP, there is generally an excessive removal of metal in regions with wide metal features (e.g., region C in FIG. 3C), whereas there may be insufficient removal of metal in wide oxide regions (e.g., see region A in FIG. 3C). The excessive thinning of metal results in high resistance, and increased circuit delays. The residual metal on top of the oxide can result in leakage between interconnects.

Thus, the conventional processes result in poor local uniformity of CMP of SiO.sub.2 isolations and other structures. As a result, the devices produced may suffer failures.

SUMMARY OF THE INVENTION

In view of the foregoing problems of the conventional methods, an object of the present invention is to provide a method for improving local uniformity of chemical mechanical polishing using a self-aligned polish rate enhancement layer.

In a first aspect of the present invention, a method of making a semiconductor device, includes forming an oxide layer on a silicon substrate, forming a nitride layer over the oxide layer, depositing one of a doped oxide layer and an undoped porous oxide layer on the nitride layer, etching trenches through the one of the doped layer and the undoped porous oxide layer, the nitride layer, and the oxide layer, depositing an undoped oxide layer to fill the trenches, and patterning the undoped oxide by chemical mechanical polishing (CMP).

With the unique and unobvious process of the present invention, uniformity of the CMP results using a self-aligned polishing rate enhancement layer.

BRIEF DESCRIPTION OF THE DRAWINGS

The foregoing and other objects, aspects and advantages will be better understood from the following detailed description of a preferred embodiment of the invention with reference to the drawings, in which:

FIGS. 1A-1D illustrate a conventional process of forming a semiconductor device;

FIGS. 2A-2D illustrate a process of forming a semiconductor device according to a first preferred embodiment of the invention;

FIG. 2E illustrates a flowchart of the process of the first embodiment;

FIGS. 3A-3C illustrate a second conventional process of forming a semiconductor device using metal CMP;

FIGS. 4A-4C illustrate a process of forming a semiconductor device according to a second preferred embodiment of the invention using metal CMP;

FIG. 4D illustrates a flowchart of the process of the second embodiment;

FIGS. 5A-5C illustrate a third conventional process of forming a semiconductor device;

FIGS. 6A-6C illustrate a process of forming a semiconductor device according to a third preferred embodiment of the invention; and

FIG. 6D illustrates a flowchart of the process of the third embodiment.

DETAILED DESCRIPTION OF PREFERRED

EMBODIMENTS OF THE INVENTION

Referring now to the drawings, and more particularly to FIGS. 2A-2D, 4A-4C, and 6A-6C, preferred embodiments of the method of forming a semiconductor device according to the invention will be described.

First Embodiment

First, it is noted that the present inventors have found that use of a self-aligned layer enhances the SiO.sub.2 polish rate above the pad nitride regions (i.e., active area). This results in a faster polish rate in these regions without altering the polish rate in regions with wide trenches, thereby resulting in a thicker SiO.sub.2 layer in the wide trenches after Chemical Mechanical Polishing.

Similarly, a self-aligned layer (e.g., preferably doped SiO.sub.2 or porous SiO.sub.2) with a fast polish rate can be placed above the pad nitride so that less over-polish is required to remove the SiO.sub.2 from regions with wide active areas. A process flow according to the invention is described hereinbelow, with reference to FIGS. 2A-2D. FIG. 2E illustrates a flowchart of the process 200 of the first embodiment. The advantages of such a method will be clear especially when comparing the inventive process with the conventional process shown in FIGS. 1A-1D.

Referring to FIG. 2A, first an insulating layer (e.g., preferably SiO.sub.2) 21, preferably having a thickness in a range of about 2 nm to about 50 nm, and more preferably about 10 nm, is formed (e.g., grown, for example by thermal oxidation) on a silicon substrate 20 (step 201).

Then, a layer 22 of Si.sub.3 N.sub.4 is formed (e.g., deposited by Low Pressure Chemical Vapor Deposition (LPCVD)) to a thickness of substantially in a range of about 50 nm to about 300 nm, and more preferably to about 100 nm (step 202).

Thereafter, a layer 23 of doped SiO.sub.2 (e.g., Boron-, Phosphorous-, or F-doped) is deposited by, for example, LPCVD to a thickness in a range of about 50 nm to about 300 nm, and more preferably about 100 nm). Alternatively to the doped SiO.sub.2, a layer 23 of porous SiO.sub.2 (e.g., preferably having a porosity substantially in a range of about 10% to about 80%) may be deposited by a spin-on-glass (SOG) process to a thickness of about 50 nm to about 300 nm, and more preferably to about 100 nm (step 203).

Then, trenches are formed (e.g., etched) through the layers and into Si substrate 20, using Fluorine or Chlorine-based Reactive Ion Etching (RIE) to a trench depth into the Si substrate 20 of about 100 to about 1,000 nm, and more preferably to about 200 nm (step 204).

As shown in FIG. 2B, undoped (fill) SiO.sub.2 24 is then formed (e.g., deposited) to fill the trenches, either by LPCVD or SOG to a thickness of about 2 times the trench depth (step 205). At this stage, a high temperature anneal (e.g., performed within a range of about 800 to about 1000.degree. C., for about 1 minute to about 60 minutes) may be used to drive impurities (e.g., Boron, Phosphorous, or Fluorine) into the undoped SiO.sub.2 24 from the underlying doped SiO.sub.2 23 (i.e., self-aligned doping of SiO.sub.2).

As shown in FIG. 2C, the fill SiO.sub.2 24 is patterned by CMP (e.g., step 206). In general, polishing is stopped when the pad nitride is exposed in all regions. However, if self-aligned doping of SiO.sub.2 is used, then the polishing can be stopped before the fill SiO.sub.2 24 is removed from the regions with wide active areas. A selective wet etch (dilute HF) can then by used to remove the doped SiO.sub.2 23.

If porous SiO.sub.2 (e.g., between about 10% to about 80% porosity) is used as layer 23 in place of doped SiO.sub.2, then the polishing is stopped when the pad nitride is exposed in all regions. It is noted that there will be relatively little thinning of the fill SiO.sub.2 24 in wide trenches during the removal of the porous SiO.sub.2 23 from the wide active areas, due to the high polish rate of porous SiO.sub.2 23 (e.g., about 5 to 10 times higher). Further, the same approach could be used when doped SiO.sub.2 forms layer 23, because it also has a higher polish rate than undoped SiO.sub.2 (e.g., about 5 to 10 times higher).

Thereafter, as shown in FIG. 2D, the doped SiO.sub.2 layer 23, if necessary, is stripped selectively to undoped SiO.sub.2 using dilute hydrofluoric (HF) acid, and the pad nitride 22 is stripped using H.sub.3 PO.sub.4 (e.g., step 207). Then, the pad oxide 21 is stripped using, for example, diluted HF, and a sacrificial SiO.sub.2 25 layer is grown as a screen oxide, for subsequent implants, using thermal oxidation (e.g., step 208).

Thereafter, conventional CMOS processing may be used to make gates, junctions, contacts, interconnects, etc. (e.g., step 209)

The process according to the first embodiment of the present invention is relatively simple, and adds only one deposition step (and optionally an extra anneal and wet etch step). The fast removal of the oxide over wide pad nitride regions results in a thicker SiO.sub.2 in wide trenches and ensures complete removal of SiO.sub.2 in the wide active areas. Thus, the use of doped SiO.sub.2 is especially advantageous in enhancing the polishing rate.

A similar method can be used to pattern polysilicon used in filling deep trenches for Dynamic Random storage capacitors. That is, the method of the first embodiment could be modified such that polysilicon fills the trenches rather than SiO.sub.2.

Thus, in this modification, the self-aligned underlayer would be doped polysilicon, whereas the material to be polished would be intrinsic polysilicon. The doped polysilicon polishes about 3 times faster than undoped polysilicon, making it easier to remove the undoped polysilicon in wide field regions.

Second Embodiment

A similar approach may be used for metal Chemical Mechanical Polishing, as shown in FIGS. 4A-4C and the flowchart of the process 400 shown in FIG. 4D.

First, as a basis for comparison a conventional metal CMP will be described referring to FIGS. 3A-3C. Referring to FIG. 3A, an interlevel dielectric (ILD) 31 is deposited and then trenches 32 are etched therein.

In FIG. 3B, a titanium/titanium nitride (Ti/TiN) liner 33 is deposited in the trenches 32 along with a conductor (e.g., tungsten) 34, thereby to fill the trenches 32.

In FIG. 3C, the conductor 34 is patterned by CMP.

However, as mentioned above, in the case of metal CMP, there is generally excessive removal of metal in regions with wide metal features (e.g., see region C in FIG. 3C), whereas there may be insufficient removal of metal in wide oxide regions (e.g., see region A in FIG. 3C). The excessive thinning of metal results in high resistance, and increased circuit delays. The residual metal on top of the dielectric 31 (see Region A in FIG. 3C) can result in leakage between interconnects. Region B in FIG. 3C represents a dense array of interconnects.

Thus, for CMP of tungsten (W) in the conventional process, the Ti liner 33 can be difficult to remove because it has a much lower polish rate than W in many slurries (e.g., acidic Al.sub.2 O.sub.3 slurries), resulting in recessing of the W in wide trenches and residual Ti in regions with wide SiO.sub.2.

A self-aligned W layer can be alloyed with the Ti to enhance the W polish rate above the wide oxide regions. This results in a faster polish rate in these regions without altering the polish rate in regions with wide trenches, thereby resulting in a thicker W layer in the wide trenches after CMP.

Similarly, a self-aligned layer with a fast polish rate can be placed above the wide oxide regions so that less over-polish is required to remove the W from regions with wide active areas. A process flow is described below with reference to FIGS. 4A-4C, and a flowchart of the method 400 of the second embodiment is shown in FIG. 4D.

First, as shown in FIG. 4A, an interlevel dielectric (e.g., SiO.sub.2 layer) 40 is formed by either LPCVD or by Plasma Chemical Vapor Deposition (PCVD), to a thickness of about 100 nm to about 1,000 nm, and more preferably to about 500 nm (step 401).

Then, a metal layer 41 (e.g., tungsten, Mo, Ta, Al, Zr, Nb, and/or Cu) is deposited by Chemical Vapor Deposition (CVD) or by Physical Vapor Deposition (PVD) to a thickness of about 10 nm to about 500 nm, and more preferably about 50 nm. Thereafter, contact holes and trenches are etched, using F-based Reactive Ion Etching (RIE) (step 402).

In FIG. 4B, a liner 42 and another metal layer 43 are formed thereon (e.g., deposited by PVD or LPCVD) (step 403). It is noted that the liner 42 may be selected depending upon the metal employed. For example, a Ta or TaN liner may be employed with copper, whereas a Ti/TiN liner is typically employed with W or Al conductors. Assuming a Ti/TiN liner, preferably, the titanium in a Ti/TiN liner has a thickness of about 5 nm to about 50 nm, and more preferably about 10 nm. Preferably, the TiN has a thickness of about 5 nm to about 50 nm, and more preferably a thickness of about 10 nm. The tungsten preferably has a thickness of about 2 times the trench depth. At this point, a high temperature anneal (e.g., preferably performed at about 400 to about 800.degree. C., for about 2 hours) may be used to alloy the Ti with the underlying metal).

Then, as shown in FIG. 4C, the tungsten is patterned by CMP (step 404). The polish is stopped when the field oxide is exposed in all regions. It is noted that there will be relatively little thinning of the W in wide trenches during the removal of the Ti from the wide active areas, due to the high polish rate TiW alloy compared to Ti (e.g., about 5 to about 10 times higher). At this point, additional metal interconnect layers could be formed if desired.

The above process shown in FIGS. 4A-4C is relatively simple, and only adds one deposition step (and optionally an extra anneal step). The fast removal of, for example, Ti of the liner 42 over wide oxide regions A results in a thicker W layer 43 in wide trenches (e.g., Region C) and ensures complete removal of liner 42 in wide active areas. Thus, in the second embodiment, using a metal underlayer enhances the polishing rate.

Additionally, extra polishes associated with removing the liner are eliminated using this approach.

Third Embodiment

According to a third embodiment of the present invention, this approach also can be used for oxide CMP over a gate stack, where the gate stack is used as a polish stop, as shown in FIGS. 6A-6C.

First, a conventional method of oxide CMP over a gate stack is shown in FIGS. 5A-5D.

First, in FIG. 5A, over a silicon substrate 50 is formed a SiO.sub.2 layer 51, a polysilicon 52 and a SiN layer 53 in turn. The gate stack is patterned, as shown.

In FIG. 5B, a dielectric 54 is deposited.

In FIG. 5C, the dielectric 54 is patterned by CMP.

To improve over the conventional process of FIGS. 5A-5C, in the third embodiment, a self-aligned layer 65 of doped SiO.sub.2 or porous SiO.sub.2 is used to enhance the SiO.sub.2 polish rate above the wide gate regions. This results in a faster polish rate in these regions without altering the polish rate in regions with wide trenches, thereby resulting in a thicker SiO.sub.2 layer in the wide trenches after CMP. A process flow is described below with reference to FIGS. 6A-6C, and a flowchart of the inventive method 600 according to the third embodiment is shown in FIG. 6D.

As shown in FIG. 6A, first, a gate SiO.sub.2 layer 61 is formed (e.g., grown by thermal oxidation), preferably having a thickness in a range of about 2 nm to about 50 nm, and more preferably 5 nm, over a silicon substrate 60 (step 601).

Then, doped polysilicon 62 (B, As, or P-doped) is formed (e.g., deposited by LPCVD) to a thickness in a range of about 20 nm to about 200 nm, and more preferably to about 50 nm (step 602).

A metal such as Ti, W, or Ta is deposited by PVD or LPCVD on the polysilicon 62. Thereafter, a silicide 62a such as TiSi.sub.2, WSi.sub.2, TaSi.sub.2 or the like, is formed (e.g., by heating of the polysilicon and the metal), to a thickness of about 10 nm to about 100 nm, and more preferably to about 50 nm (step 603). The preferred silicide is WSi.sub.2 because of high thermal stability and low resistivity.

Then, Si.sub.3 N.sub.4 63 is formed (e.g., preferably deposited by LPCVD) to a thickness of about 50 nm to about 300 nm, and more preferably to about 150 nm (step 604).

Doped SiO.sub.2 64 (e.g., B-, P-, or F-doped) is formed (e.g., deposited by LPCVD) to a thickness of about 50 nm to about 300 nm, and more preferably to about 100 nm. Alternatively to doped SiO.sub.2, porous SiO.sub.2 may be deposited by SOG to a thickness of about 50 nm to about 300 nm, and more preferably to about 100 nm.

Thereafter, the stack is etched using F and Cl-based Reactive Ion Etching (RIE), and junctions, silicides, etc. are formed (e.g., step 605).

Referring to FIG. 6B, doped (e.g., doped with B, P, or F) or undoped SiO.sub.2 65 is formed (e.g., preferably deposited by LPCVD) over the gate stack to a thickness of about 2 times the trench depth (step 606). At this point, a high temperature anneal may be used to drive additional impurities (e.g., B, P, or F) into the fill SiO.sub.2 65 from the underlying doped SiO.sub.2 64 (i.e., self-aligned doping of SiO.sub.2).

Referring to FIG. 6C, the fill SiO.sub.2 65 is patterned by CMP (step 607). If self-aligned doping of SiO.sub.2 is used, the polishing can be stopped before the fill SiO.sub.2 65 is removed from the regions with wide active areas. A selective wet etch can then by used to remove the doped SiO.sub.2 64.

If porous SiO.sub.2 is used as layer 64, then the polish is stopped when the nitride 63 is exposed in all regions. It is noted that there will be relatively little thinning of the SiO.sub.2 65 in wide trenches during the removal of the porous SiO.sub.2 64 from the wide active areas, due to the high polishing rate of porous SiO.sub.2 64 (e.g., about 5 to 10 times higher). Furthermore, it is noted that the same approach could be used when doped SiO.sub.2 forms layer 64, because it also has a higher polish rate than undoped SiO.sub.2 (e.g., about 5 to 10 times higher).

At this point, conventional CMOS processing would be used to make contacts, interconnects, etc. (step 608).

Thus, with the third embodiment, an oxide CMP over the gates with doped SiO.sub.2 is used to enhance the polishing rate.

Moreover, with each of the above-described embodiments, enhanced local uniformity of the CMP results from using a self-aligned polishing rate enhancement layer. The enhanced uniformity of the dielectric improves the yield for contacts to Si. That is, too much over-etching of the contacts damages Si, whereas too little over-etching results in high resistance. If the oxide thickness varies, then it is impossible to optimize the over-etch.

While the invention has been described in terms of several preferred embodiments, those skilled in the art will recognize that the invention can be practiced with modification within the spirit and scope of the appended claims.

Top

Current U.S. Class:	438/359; 257/E21.244; 257/E21.548; 438/360
Intern'l Class:	H01L 021/331
Field of Search:	438/359,361